BROWSE
Detail
Conference Paper
Solar-powered speed display system using image-based speed detection on an edge AI camera
Cited 0 time in 
Share 
Share
- Authors
- Jang Woon Baek, Yun Won Choi, Jinhong Kim, Jaejin Lee
- Issue Date
- 2025-02
- Citation
- International Conference on Artificial Intelligence in Information and Communication (ICAIIC) 2025, pp.1098-1100
- Language
- English
- Type
- Conference Paper
- Abstract
- This paper proposes a solar-powered speed display system using camera-based speed detection on low-power, cost-effective edge devices. Recently, solar-based optical fiber traffic signs with high visibility during both day and night have gained popularity. Concurrently, there is a growing interest in intelligent traffic services powered by low-energy edge devices integrated into traffic sign systems. Our system utilizes a Raspberry Pi 5 as an edge device to overcome limitations in existing radar-based systems, which struggle to distinguish vehicles from pedestrians. The Raspberry Pi 5, selected for its affordability and low power consumption compared to alternatives like NVIDIA Orin Nano, receives video input to detect vehicle speed using a lightweight YOLO-fastest object detector and a high-speed IoU tracker. Through simple calibration and distance mapping between real road distances and image pixels, the system achieves rapid and accurate speed detection. Real-world tests demonstrate a high accuracy rate exceeding 95%, validating the effectiveness of the proposed system.
- KSP Keywords
- Accuracy Rate, Display System, Distance mapping, Edge devices, High-speed, Image-based, Intelligent traffic, Low energy, Low power consumption, Optical fiber, Raspberry PI